when it had only two comments. One of them was the Gemini summary, which had already been massively downvoted. I couldn't make heads or tails of the paper posted, and probably neither could 99% of other HNers. I was extremely happy to see a short AI summary. I was on my phone and it's not easy to paste a PDF into an LLM.
When something highly technical is posted to HN that most people don't have the background to interpret, a summary can be extremely valuable, and almost nobody is posting human-written summaries together with their links.
If I ask someone a question in the comments, yes it seems rude for someone to paste back an LLM answer. But for something dense and technical, an LLM summary of the post can be extremely helpful. Often just as helpful as the https://archive.today... links that are frequently the top comment.
https://news.ycombinator.com/item?id=46204895
when it had only two comments. One of them was the Gemini summary, which had already been massively downvoted. I couldn't make heads or tails of the paper posted, and probably neither could 99% of other HNers. I was extremely happy to see a short AI summary. I was on my phone and it's not easy to paste a PDF into an LLM.
When something highly technical is posted to HN that most people don't have the background to interpret, a summary can be extremely valuable, and almost nobody is posting human-written summaries together with their links.
If I ask someone a question in the comments, yes it seems rude for someone to paste back an LLM answer. But for something dense and technical, an LLM summary of the post can be extremely helpful. Often just as helpful as the https://archive.today... links that are frequently the top comment.